Semantic Output
- The first set of results is the semantic results.
- The test has 27 sentences, and each time the parser gets the
correct results.
- To get the results, all activation is stopped and the
verb instance is turned on.
- Via binding, activation spreads to the access subnets and
to the other instances.
- It's measured after 45 cycles.
- Activation can then be counted to see what is active and
generate a symbolic output.
- This can also be restricted to see what, for example,
is the object of the sentence.
- From an NL engineering perspective, this is not very interesting.
- It merely shows that it can be made to work on a wide range of
sentences.
- (I'm pretty sure I can make it work for any given sentence. Try me.)